trainer_state.json

{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 2.0,
  "eval_steps": 500,
  "global_step": 12,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.17,
      "grad_norm": 5.802258301213437,
      "learning_rate": 1e-05,
      "loss": 3.714,
      "step": 1
    },
    {
      "epoch": 0.83,
      "grad_norm": 4.228032304421419,
      "learning_rate": 1.5877852522924733e-05,
      "loss": 3.6492,
      "step": 5
    },
    {
      "epoch": 1.0,
      "eval_loss": 5.712366104125977,
      "eval_runtime": 0.0483,
      "eval_samples_per_second": 248.688,
      "eval_steps_per_second": 20.724,
      "step": 6
    },
    {
      "epoch": 1.67,
      "grad_norm": 2.84117612385776,
      "learning_rate": 1.9098300562505266e-06,
      "loss": 3.457,
      "step": 10
    },
    {
      "epoch": 2.0,
      "eval_loss": 5.508264064788818,
      "eval_runtime": 0.0486,
      "eval_samples_per_second": 247.068,
      "eval_steps_per_second": 20.589,
      "step": 12
    },
    {
      "epoch": 2.0,
      "step": 12,
      "total_flos": 36238786560.0,
      "train_loss": 3.5401106079419455,
      "train_runtime": 10.4273,
      "train_samples_per_second": 141.744,
      "train_steps_per_second": 1.151
    }
  ],
  "logging_steps": 5,
  "max_steps": 12,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 2,
  "save_steps": 1000000000,
  "total_flos": 36238786560.0,
  "train_batch_size": 32,
  "trial_name": null,
  "trial_params": null
}